1

The Death of a King and Volatility in the Jordanian Capital Market

Aktham Maghyereh &

Ghassan Omet

Faculty of Economics & Administrative Sciences The Hashemite University

Jordan

Abstract The Late His Majesty King Hussein of Jordan, the region’s longest-serving leader, died on Sunday, 7 February 1999. Although it came as no surprise, that did not lessen the grief by Jordanians. Indeed, this event in the contemporary history of Jordan is important for its emotional, political and economic implications. This paper uses the standard GARCH, exponential GARCH (EGARCH) and the GJR models to examine the conditional volatility associated with two major dates associated with King Hussein. These are the official denial that Hussein may die within three months and the date of his death. The results indicate that King Hussein’s health “rumors” and the date of his death caused extremely high volatility in the Jordanian capital market. Moreover, the results suggest that these events are discounted in at least three days following their respective dates. Correspondence to: Ghassan Omet Dean College of Economics & Administrative Sciences PO Box, 150459 The Hashemite University, Zarqa, Jordan. Tel: 05 3826600; Fax: 05 3826613 E-mail: gomet@hu.edu.jo

2

I. Introduction

The Hashemite Kingdom of Jordan occupies an area of about 89,000 square

kilometers. Its population is about 4.5 million and with an annual increase of about

3.0%, Jordan is considered to be one of the fastest growing countries in the world.

The demographic profile of Jordan is similar to that in most advanced economies. The

urbanization rate is about 70%, life expectancy is around 70 years and the adult

literacy rate is about 85%.

Jordan’s natural resources base is very limited. Other than Potash and Phosphate, the

country has no natural resources. This is why, traditionally, the economy has relied

heavily on foreign aid, worker’s remittances and external debt.

Any attempt to analyze the economic performance of Jordan is fraught with numerous

difficulties. The country has passed through many internal and external disturbances

(economic and political) and these make the detection of any underlying economic

trends extremely difficult. These disturbance sources include the 1967 Arab Israeli

War, the 1970 Civil War, the 1973 War, the 1989 devaluation of the Jordanian Dinar

and the introduction of the IMF supported austerity measures, the 1991 Gulf War, the

1994 signing of the peace treaty with Israel, the 1998 beginning of the late His

Majesty King Hussein's (HMKH) chemotherapy for non-Hodgkin's lymphoma, the

1999 death of HMKH, and more recently the 2001/2002 Intifada in the Palestinian

territories.

3

The Hashemite Kingdom of Jordan (HKJ) was blessed with a pivotal figure "The Late

His Majesty King Hussein". Hussein ibn Talal was only a teenager when he assumed

the throne of the HKJ in 1952. At the age of 15, he witnessed the assassination of his

grandfather. He inherited a poor Kingdom, unstable and awash in refugees from the

1948 Arab-Israeli War. Moreover, Hussein's early life was complicated. He survived

a number of assassination attempts, his authority was challenged by his own prime

ministers, threats from other Arab countries and occasional riots by Arab nationalists.

Within a few years, Palestinian guerilla movements challenged the King's authority

and after the break-out of the 1973 Arab-Israeli War, Hussein had to face the invasion

of Kuwait by Iraq. Despite of these "disturbances", the late King Hussein was known

as a force for peace and stability and in 1994, he signed a peace treaty with Israeli

Prime Minister Yitzhak Rabin.

The Amman Securities Market (ASM) was established in 1978. Since its formation,

the market has experienced some growth in a number of aspects. In 1978, for

example, the total number of listed companies was 66. By the end of 2002, this

number increased to 161. Moreover, the fact that the market capitalization of all listed

companies as a proportion of GDP is equal to about 76% (2002) is an indication of

the relative importance of the market. However, similar to many emerging markets,

the ASM suffers from a number of weaknesses. In 2002, for example, 10 companies

only accounted for about 70% of the total market in terms of market capitalization

and trading volume. In other words, the market is concentrated in terms of market

capitalization and trading volume and most listed shares are thinly traded on the

secondary market.

4

Relative to the above-mentioned political and economic instability sources, it can be

argued that the most natural occurrence and peaceful change that Jordan had to face,

the death of King Hussein, was, potentially, the most important as far as its impact on

ASM’s share prices. Indeed, this event in the contemporary history of Jordan is

important for its emotional, political and economic implications. King Hussein's

death was felt far outside the grieving Kingdom. At the funeral, former presidents

Clinton, Bush, Carter, and Ford represented the United States. The British prime

minister, the French president, and the German Chancellor were also present. The

former president of Russia (Yeltsin), despite of his illness, also attended the funeral.

In a show of unusual political diversity, leaders of most Arab leaders were side by

side with many western political leaders.

Given the special status of the late His Majesty King of Hussein, this paper attempts

to examine the impact of his death on the conditional volatility of the ASM. In Table

1, we report the events surrounding the date of Hussein's death.

Table 1 Highlights of King's Last Days

July 1998 Begins chemotherapy for non-Hodgkin's lymphoma September 21, 1998 King Hussein assures his people about his health. October 1998 Helps stave off the collapse of the Israeli-Palestinian peace accord

by leaving hospital and appearing at the Wye River peace talks. October 20, 1998 Jordan categorically denied a report by the British "Foreign

Report" intelligence bulletin that King Hussein is in his last days and may die within three months.

January 19, 1999 Returns to a tumultuous welcome in Jordan after cancer treatment.

January 25, 1999 Appoints his eldest son, Prince Abdullah, as heir and hours later, he returns to the USA for further treatment.

February 4, 1999 Hussein's private physician announces that the King's treatment failed and king return home.

February 7, 1999 King Hussein dies

5

The issue of stock market volatility has received great attention in the finance

literature. High levels of volatility can adversely affect stock markets and undermine

the financial system as a whole. Volatility also discourages risk-averse investors and

savers, and stock market fluctuations may raise the cost of capital to corporations1

and may also increase the value of the “option wait” and hence delay investments.

Volatility tends also to discourage firms from seeking a stock market listing or

attempting to raise funds by new issues. Thus, high levels of market volatility can

impede investment and slow overall economic growth (De Long et al., 1989) 2.

A large number of theoretical and empirical studies have attempted to explain the

sources of volatility in stock market returns3. The debate surrounding non-

fundamentals in equity markets has been rekindled in recent years by the publication

of the volatility tests of Shiller (1981) and LeRoy and Porter (1981). They both found,

based on a simple present value model with a constant discount rate, that stock

market volatility is greater than could be justified by subsequent changes in

dividends. These studies were then followed by Roll (1988) who found that only

approximately one-third of the monthly variation in individual stock returns can be

explained by systematic economic influences. The results of Cutler et al. (1989)

indicate that macroeconomic news can explain only between one-fifth and one-third

of the movements in stock market indices. These conclusions can also be found in

Schwert (1989) who found that, although there is weak evidence that macroeconomic

1 See for example Bekaert and Wu (2000), Singh (1992, 1997, 1999), Glen, et al., (2000), Singh and Weisse (2000) and Kim and Singal (2000). 2 Bittlingmayer (1998) finds that stock market volatility had a strongly negative effect on investment and industrial production in Germany during the interwar period. 3 See for example, Flood and Graber (1984), Flood and Hodrick (1986), De Long et al. (1989, 1990) and Diba and Grossman (1988).

6

volatility can help predict stock return volatility, the amplitude of the fluctuations in

aggregate stock market volatility is difficult to account for using simple models of

stock valuation, especially during the Great Depression.

In addition to the impact of economic variables, it is plausible to argue that non-

economic variables could contribute to stock market volatility. While there have been

many papers that examined the impact of political events on business cycles4, few

papers have examined the impact of political events on financial markets. This may

due to the lack of data or difficultly of modeling political events. Among the limited

studies are Willard et al. (1996) who found some significant evidence that the turning

points in the US Civil War were reflected in the price of the Greenbacks. Erb et al.

(1996) also found a significant relationship between equity market volatility and

political risk as measured by the ICRG political risk ratings in emerging markets.

Recent work by Bittlingmayer (1998) suggests a close relationship between political

risk and market volatility during the transition from the Imperial to Weimar Germany.

However, Cutler et al. (1989) found little evidence of political news having a

significant impact on the US market.

This paper attempts to add to this “growing” literature by focusing on the explanatory

power of the sad and somewhat expected death of King Hussein on the volatility of

equity returns in the Jordanian capital market. The analysis of volatility in leading

finance journals has been advanced by the development of sophisticated econometric

models in which the conditional variance of returns is permitted to vary over time.

4 See, among others, Alesina (1987), Alesina and Cukierman (1990), Alesina et al. (1996), Alesina and Perotti, 1996, Caselli et al. (1996) and Campos et al. (1999).

7

The principle advantage of employing such models is the ability to capture the

common empirical observations in daily time series: fat tails due to time-varying

volatility, skewness resulting from mean non-stationarity, non-linearity dependence,

and volatility clustering.

In this paper, we employ three different Autoregressive Conditional

Heteroscedasticity (ARCH) specifications to examine the impact of the death of King

Hussein on the volatility in the ASE. Specifically, we use the standard GARCH,

exponential GARCH (EGARCH) and the GJR models. Then, we employ various

specification tests to determine which model provides an adequate explanation of

returns and volatility in the Jordanian market during the period under investigation.

The empirical findings have important implications for market participants as well as

policy makers in Jordan and other emerging markets. An empirical investigation of

how markets process new information about political news may broaden our general

understanding of equity return volatility, and help investors price such risks

effectively. For policymakers, the evidence indicates that political news has strong

implications for stock market volatility. In turn, stock market volatility is known to

have growth implications in emerging economies5. Highly volatile market returns due

to political news may scare investors, i.e. especially foreign investors.

The remainder of the paper is organized as follows. In Section II, we provide a

description of the methodology used in the paper. In Section III, some statistical

5 Levine and Zervos (1998) and Spyrou and Kassimatis (2001) reported that stock market development measured by volatility of stock returns has negative effect on economic growth.

8

analyses are carried out to verify that the often-cited properties of daily financial

time-series data are represented in our sample. Section IV contains a description of

the model and a discussion of the empirical evidence. Finally, section V concludes

the paper.

II. Methodology

Prior to estimating the impact of the somewhat expected death of King Hussein on the

volatility of the ASM, we need to determine an appropriate model that can best

describe the behaviour of stock returns during the period under consideration. In this

study, we employ three different Autoregressive Conditional Heteroscedasticity

(ARCH) specifications: the standard GARCH, exponential GARCH (EGARCH) and

the GJR models. We employ various specification tests to determine which model

provides an adequate explanation of returns and volatility in the Jordanian market.

Finally, the appropriate model is used to calculate the conditional volatility of stock

returns over time i.e., time-varying volatility.

The GARCH class of models used in this study has proven to be particularly suited

for modeling the behavior of financial data. As emphasized by Pagan (1996), these

models are capable of capturing the common characteristics of many financial time

series. First, asset prices are generally non-stationary and often have a unit root,

whereas returns are usually stationary. Second, returns series usually show little

autocorrelation, while serial independence between the squared values of the series is

often rejected pointing towards the existence of non-linear relationships between the

subsequent observations. Volatility of returns appears to be clustered. Returns go

9

through periods of high and low variances. These facts point towards time-varying

conditional variances. Most empirical evidence indicates that the empirical

distribution of returns differs significantly from the identical Gaussian distribution.

The series are characterized by leptokurtosis, which could be related to the time-

variation in the conditional variance. Finally, some series exhibit asymmetric

behavior in the conditional variance (leverage effects).

The autoregressive Conditional Hetroscedasticity model (ARCH) has been introduced

by Engle (1982) and generalized by Bollerslev (1986). The specification of the

conditional variance in a ( )qpGARCH , model is given in Equation 1 below:

( )2

1

2

1

21

1

jt

p

jjit

q

iit

tt

hh

uh

−=

−=

∑∑ ++=

=

βεαω

ε

with tu being IID 10,0,0,1

<+≥≥> ∑∑=

q

jj

p

iiji and βαβαω .

In practice, numerous studies have demonstrated that the ( )1,1GARCH specification is

the most appropriate. In this model, the conditional variance is a function of three

terms: First, the mean, ω . Second, news about volatility from the previous period,

measured by the lag of the squared residual from the mean equation, 21−tε (the ARCH

term). Third, last period’s forecast variance, 21−th the (GARCH term). The coefficients

of the model are easily interpreted. The estimate of 1α shows the impact of current

news on the conditional variance process and the estimate of 1β the persistence of

10

volatility to a shock or, alternatively, the impact of “old” news on volatility. Engle

and Bolleslev (1986) show that the persistence of shocks to volatility depends on the

sum of βα + . Values of the sum lower than unity imply a tendency for the volatility

response to decay over time, and at a slower rate the closer the sum is to unity. In

contrast, values of the sum equal to (or greater) unity imply indefinite (or increasing)

volatility persistence to shocks over time.

It is often observed that downward volatilities in financial markets are followed by

higher volatilities than upward movements of the same magnitude. However, the

GARCH model imposes symmetry on the conditional variance structure that may not

be appropriate for modeling the behavior of stock returns. To deal with this problem,

Nelson (1991) proposes the exponential GARCH or EGARCH model. Under the

EGARCH (1,1) the conditional variance is given by:

( )2)log(2)log(1

11

1

1

−

−−

−

− ++

−+=

t

tt

t

tt h

hh

hεδβπεαω

where δβαω and,,, are parameters.

The EGARCH model has three distinct advantages over the GARCH model. First, the

logarithmic construction of equation 2 ensures that the conditional variance is strictly

positive, thus, the non-negative constraints used in the estimation of the GARCH are

not necessary. Second, since the parameter δ typically enters Equation 2 with a

negative sign, bad news, 0<tε , generates more volatility than good news. Finally, the

11

degree of persistence in this model is measured only by the δ parameter: if 1<δ , the

model is not integrated.

A frequently used alternative specification to the conditional volatility process is the

model proposed by Glosten et al. (1993). This model (the GJR) extends the GARCH

model to allow for asymmetric effects by including an indicative dummy as shown by

the following equation:

( )32111

21 −−−− +++= ttttt dhh εδβεαω

where 1−td is a dummy variable that takes the value of unity if 01 <−tε and zero

otherwise.

The GJR model is closely related to the threshold GARCH, or TGARCH model of

Rabemananjara and Zakoian (1993) and Zakoian (1994). According to the GJR

model, if bad news has greater impact on volatility than good news, a leverage effect

exists, and we expect 0>δ . The impact of good news will be α while bad news has

an impact of δα + . The β parameter measures the degree of persistent in the

conditional variance. The sum of the parameter values of δβα and,, measures the

persistence in volatility shocks. If the sum of these parameters is less than one, the

shock dies out over time; a value close to one means that a shock will affect the

conditional variance and the forecast of it for quite some time. If the sum of the

parameters is equal to one, the shock will affect volatility into the indefinite future.

12

Models with the sum of these parameter values that equal one are called integrated

GARCH models or IGARCH.

To complete the GARCH model, a mean equation must be specified. Thus, we

propose the following specification for the mean equation:

( )4,...,11 TtLogPLogP ttt =++= − εµ

where, tP and 1−tP are the stock prices at time t and t-1 respectively, and tε is a

random error (independent and normal distribution, ),0(~ 2σε INDt ).

If Equation 4 is re-arranged so that the lagged logarithmic stock price is moved to the

left-hand side of the equal sign, we get a function that states that relative stock returns

will be a linear function of a constant and news (residuals) - the mean equation. In the

empirical test, we will add a moving average term, MA(1), on the right-hand side of

the mean equation to cope with any serial correlation which may be caused by non-

synchronous trading in the stocks.

The main assumption under the GARCH models mentioned above is that the

conditional distribution of returns is normal, i.e. the standardized residuals of these

models should be normal. However, in practice, despite the flexible functional forms

specified above, there is often excess kurtosis in the standardized residuals of

GARCH models. To deal with this problem, Bollerslev and Wooldrige (1992) studied

quasi-maximum likelihood (QML) estimation of GARCH models. They argued that

13

under a correct specification of the first and second moment, consistent estimates of

the parameters of the model could be obtained by maximizing a likelihood function

constructed under the assumption of conditional normality. Consistent parameter

estimates can then be obtained using a robust covariance matrix estimator. Therefore,

in this study we compute the QML covariance and standard errors using the methods

described by Bollerslev and Wooldrige (1992) and Glosten et al. (1993).

To determine which specification provides an adequate explanation of returns and

volatility in the Jordanian market, some diagnostic tests are performed. First, we

examine whether the standardized residuals of the estimated models display excess

skewness and kurtosis. If properly specified, the GARCH models should be able to

significantly reduce the excess skewness and kurtosis present in normal returns.

Second, we examine whether the squared standardized residuals of the models are

independent and identically distributed. Third, we examine whether the standardized

residuals of the estimated models display ARCH effects (ARCH-LM test). Properly

specified models should be able to significantly remove the ARCH effects. Fourth,

we test whether the standardized residuals of the estimated models display non-

linearity relationships (Ramsey (1969) REST test). The Schwartz Information

Criterion (SIC) is also reported for each specification used, which allows degrees of

freedom free comparison of the models’ performance.

III. Data Description

To investigate the effect of the death of King Hussain on the volatility of the ASM,

the daily closing price index of the market for the period from January 1998 to

14

December 1999 is used. The daily stock returns, rt, are calculated using the first

differences of the logarithmic price index, i.e. )()( 1−−= ttt PLnPLnr where P is the

price index in days t and t-1 respectively. After excluding non-trading days, the daily

time series consists of 486 observations.

The daily logarithmic returns and squared returns are first tested for the presence of

autocorrelation and stationarity. Table 2 reports descriptive statistics which show that

the returns exhibit skewness and significant kurtosis.

The series indicates significantly flatter tails than does the stationary normal

distribution. The coefficients of skewness indicate that the series typically has

asymmetric distributions skewed to the right. In addition, the return series displays

excess kurtosis. In this case, the null hypothesis of coefficients conforming to the

normal value of three is rejected. Thus, the return series is leptokurtic. That is, its

distribution has thicker (flatter) tails than the normal distribution. The hypothesis of

normality is rejected by the bivariate Jarque-Bera test and this confirms the results

based on either skewness or kurtosis. From the Ljung-Box test statistics for twelfth-

order serial correlation for the levels, the returns show autocorrelation. We thus reject

the null hypothesis of no serial correlation and homoscedastic daily returns

(uncorrelated squared returns). Furthermore, the ARCH-LM test statistics report

highly significant changing volatility. These results suggest the presence of time-

varying volatility. This is the so-called stylized fact of volatility clustering in return

series.

15

Table 2 Stock Market Returns: Some Basic Statistics

Mean -0.00001

Standard Deviation 0.00649 Kurtosis 5.1980

Skewness 0.3673 Q(12) 70.520*

Q2 (12) 78.766* Normality 103.628*

ADF -9.784* PP -15.868

ARCH(6)-LM 11.072* REST(12;6) 12.352*

Normality represents the Jarque-Bera (1980) normality test, which follows a chi-squared distribution, with two degrees of freedom. Q(2) and Q2(12) are Ljung-Box tests for serial correlation in the returns and squared returns data respectively. ADF and PP are Augmented Dickey Fuller and Phillips-Perron tests for a unit root, respectively. ARCH (6)-LM is a test for conditional heteroscedasticity in returns for sixth order. The OLS-regression

.... 266

2110

2−− ∗++∗+= tt yyy ααα 2RT ∗ is used and is 2χ distribute with 6 degrees of freedom. T is the

number of observations, y is returns and R2 is the explained over total variation. 610 ..., ααα are parameters. REST (12,6) is a test for non-linear dependence linearity in conditional mean of return: 12 is the number of lags and 6 is the number of moments that is chosen in the implementation of the test statistic. 2RT ∗ distributed with 12 degrees of freedom. * Significant at the 1 percent level.

Table (2) also shows that the Phillips-Perron (PP) and the Augmented Dickey Fuller

(ADF) unit root tests strongly reject the hypothesis of non-stationarity, indicating that

the return series displays a degree of time dependence. These results support the

behavior of stock prices being characterized by a martingale process. The REST

(Ramsey, 1969) test statistic suggests the presence of non-linearity in the series.

Based on the above, the ASM stock returns tend to be characterized by positive

skewness, excess kurtosis and deviation from normality. These findings are consistent

with the findings from other emerging markets6. The results also display a degree of

6 Bekaert et al. (1998) provide evidence that 17 out of the 20 emerging countries examined (the sample includes Jordan) had positive skewness and 19 out of 20 have excess kurtosis, so that normality was rejected in more than half of the countries.

16

volatility clustering7 and non-linear dependence in returns. Following Diebold (1986),

these characteristics suggest that the GARCH specification provides a good

approximation that captures the time-series characteristics of the daily returns in the

Jordanian stock market during the period under consideration.

IV. The Empirical Results

Prior to estimating the impact of King Hussein’s health “rumors” and the period

surrounding the date of his death on the volatility of ASM, we need to determine the

appropriate model that can best describe the behavior of stock returns during the

period under consideration. It is well-known that the GARCH-type models provide an

appropriate model for stock returns. Previous studies have reported that a GARCH

(1,1) specification captures the conditional volatility of the returns quite well. In this

paper, as discussed earlier, we employ three different conditional variance

specifications: the standard GARCH, EGARCH, and the GJR models, in order to

examine the sensitivity of the results to employing a variety of volatility

specifications.

Table 3 reports the results of the different GARCH specifications for the stock market

returns during the period of January-1998 to December-1999.

7Bekaert and Harvey (1997) found that most emerging stock markets including Jordan exhibit time-varying volatility.

17

Table 3

Estimates of Volatility Models Mean Equation: tt MAR εγµ ++= )1( Variance Specifications: Standard GARCH (1,1): 2

12

1 −− ++= ttt hh βαεω EGARCH(1,1):

1

11

1

1 )log(2)log(−

−−

−

− ++

−+=

t

tt

t

tt h

hh

hεδβπεαω

GJR: 2

1112

1 −−−− +++= ttttt dhh εδβεαω GARCH EGARCH GJR µ -0.00041

(-1.5042) -0.00003 (-0.8931)

-0.0003 (-1.1769)

γ 0.2741* (5.9423)

0.2743* (4.8094)

0.2587* (5.1965)

ω 0.00008* (2.8288)

0.00009* (3.9248)

-4.7318**

(-2.3076) α 0.26319*

(3.7561) 0.35711* (3.5882)

0.4043*

(3.4988) β 0.53236*

(5.0433) 0.49784* (5.0108)

0.5761* (2.9569)

δ -- -0.18175**

(-2.0864) -0.12585 (-1.4225)

Log likelihood 1817.140 1818.738 1813.087 SIC 1810.4242 1810.6781 1805.0271 Q(12) 18.659*** 19.310*** 20.502** Q2(12) 14.919 11.340 22.739** Skewness 0.2408 0.1848 0.1942 Kurtosis 3.6166 3.5731 3.8558 Jarque-Bera 39.333* 32.961* 36.629* ARCH(6)-LM 1.9329*** 1.3248 3.0727* REST(12,6) 6.6538** 3.2149 5.9672**

Bollerslev and Wooldridge (1992) quasimaximum likelood t values in brackets. SIC is the Schwartz Information Criterion: SIC=L-0.5 P*log(T), where P is the number of estimated parameters. See Table 1 for definition of the Jarque-Bera, Q(12), Q2(12), ARCH(6)-LM and REST(12,6) test statistics. *, **, *** significant at 1%, 5% and 10% respectively

The mean equation in all specifications includes an MA(1) term in order to remove

any serial correlation in the returns which may be caused by non-synchronous trading

in the stocks. Starting with the mean equation results, we observe that the MA(1) term

is highly significant in all specifications. This indicates that the returns exhibit serial

18

correlation. In other words even after we take into consideration the impact of non-

synchronous trading, the market reflects inefficiency during the period of the study.

The estimated βα and parameters in the conditional variance equation for the

model with GARCH(1,1) specification show significant ARCH and GARCH with

βα + less than unity. This implies that the effect of shocks on volatility tends to

decay within a few time lags (the duration of a shock to volatility is estimated to be

less than one month) 8. When the conditional variance equation is specified as an

EGARCH(1,1) or GJR(1,1) almost similar results to the GARCH(1,1) specification

are found. The estimated β parameter in EGARCH(1,1) is around 0.50, indicating a

tendency for the volatility response to shocks to display a shorter memory. A leverage

effect is also detected in this model - a significant negative value of the δ parameter.

Similarly, the GJR specification results provide a clear evidence of asymmetry in the

returns, as indicated by the statistically significant coefficient of δ -bad news has a

greater impact on conditional volatility than good news. Thus, the asymmetry impact

of shocks on volatility is captured by both models.

Turning to the specification statistical tests, the bottom of Table 3 shows that the

EGARCH model has the highest log-likelihood value and thus the best model. The

SIC also ranks this model first followed by the GARCH and GJR models. Although

the log-likelihood and SIC slightly favour the EGARCH, all other specification tests

(the diagnostics statistical tests) also suggest that this model provides a better fit and

captures most of the time series characteristics of the returns during our sample

8 See Lamoueux and Lastrapes (1990).

19

period. As a first diagnostic test, the twelfth order Lung-Box is calculated for the

standardized residuals9 Q(12) and squared standardized residuals Q2(12) for all the

three models. For all the models, Table 3 shows significant evidence of serial

correlation in the standardized residuals (Q(12)) at the 10 percent level. However, for

the squared residuals up to lag 12 (Q2(12)), the EGARCH reports no significant

evidence of autocorrelation. The values of kurtosis and skewness for the standardized

residuals have lower values in all models. Hence, the EGARCH suggests clearly more

normal residuals. While the ARCH-LM (6) test statistics report conditional

hetroscedasticity for the GARCH and GJR, the homosdacticity is detected in the

EGARCH. Although the REST test statistics rejected linearity in the mean for the

GARCH and GJR, linearity cannot reject in the EGARCH.

Since, all the specification tests suggest that the EGARCH model seems to capture

most of the market dynamics appropriately during the sample period of this study, we

report the conditional volatility of the ASE based on this model. Figure 1 plots the

conditional volatility of the ASE for the period from January 1998 to December 1999.

As can seen, volatility shows a cyclical pattern with significant volatility during May-

October 1998 and January-March 1999. This provides some evidence about the

significant effect of King Hussein’s health “rumors” (October 1998) and the period

surrounding the date of his death (February 1999).

9 Standardized residuals are calculated as ( )2// −∗

iitt h υυε where υ is the degree of freedom in the student-t distribution.

20

0,00E+00 2,00E-05 4,00E-05 6,00E-05 8,00E-05 1,00E-04 1,20E-04 1,40E-04 1,60E-04 1,80E-04 2,00E-04

3-1-98 8-2-98 11-3-98 19-4-98 24-5-98 28-6-981-8-982-9-984-10-984-11-9812-12-9813-1-9924-2-991-4-994-5-997-6-99 13-7-99 17-8-99 19-9-9920-10-9923-11-9926-12-

Fig . 1 Estimated Conditional Volatility of the Stock Returns

Table 4 Conditional Volatility Surrounding King Hussein’s Last Days

Trading Days Jordan categorically denied that

King Hussein may die within three months.

Death of King Hussein

October 20 1998 February 7 1999 -5 -27.608 -20.304 -4 -18.587 -19.919 -3 -6.6404 37.779 -2 -13.567 9.5244 -1 -11.019 21.049 0 124.261 161.373 +1 19.119 69.528 +2 14.079 24.140 +3 26.524 39.887 +4 -16.074 32.937 +5 -12.483 -28.394

Moreover, Table 4 reports the time varying percentage change in volatility

surrounding the two dates: rumor denial and death of the King. Clearly, we can see

that after three trading days, following the government’s denial that King Hussein

may die, conditional volatility started to decrease. Similarly, it took the market four

trading days after the death of King Hussein to reduce its volatility.

21

For a more formal test of the impact of King Hussein’s health “rumors” and the

period surrounding the date of his death on the volatility of ASE, we use two further

tests. First, we regress the time-paths of the conditional volatility (V) obtained from

the above EGARCH specification on the current and five-lagged dummy variable (D)

which takes a value of 1 when there is a main event and 0 otherwise. Second, we re-

estimate the EGARCH model after adding the current and five-lagged dummy

variable (D) mentioned above to the conditional volatility equation. The current and

lagged values of up to five days are used in both tests to allow the market to

completely absorb the effects of such events. The dates of these events are reported in

Table 1. The estimated equations are given as follows:

( )55

0tit

iit DV εηρ ++= −

=∑

( )6)log(2)log(5

01

11

1

1it

ii

t

tt

t

tt D

hh

hh −

=−

−−

−

− ∑+++

−+= ηεδβπεαω

The results of these two tests are reported in Table 5. As can be seen, the results lend

qualified support to the above evidence that suggests that King Hussein’s health

“rumors” and the period surrounding the date of his death were the prime culprits

behind the extremely high volatility of the ASE during the period of interest.

Moreover, the results of both tests indicate that the events are discounted in at least

three days following the dates of the respective events. To explain these results, one

can rely on either of the following two arguments. First, there is a high level of

information asymmetry in the market that makes participants take a long time to

arrive at a new equilibrium. Second, given the fact that the loss of King Hussein was

22

a big event, it is natural for the market to take a number of days to completely adjust

it self to a new level.

Table 5.

The Impact of Dates Surrounding King’s Death OLS Estimation:

titi

it DV εηρ ++= −=∑

5

0

EGARCH(1,1): Mean Equation: tt MAR εγµ ++= )1( Variance Specifications:

1

5

01

11

1

1 )log(2)log( −=−

−−

−

− ∑+++

−+= t

ii

t

tt

t

tt D

hh

hh ηεδβπεαω

OLS EGARCH µ 0.000027*

(5.96122) -0.00056** (-2.50045)

γ

0.24778* (4.7148)

ω 0.00078* (3.7853)

α 0.15711* (2.3483)

β 0.39617* (6.6583)

δ

-0.06131** (--1.9598)

0η 0.000289* (4.55852)

0.79310* (5.53755)

1η 0.000239* (4.28462)

0.66626* (4.44484)

2η 0.000205** (2.574139)

0.42755* (2.81878)

3η 0.0000744** (2.098954)

0.256383** (2.12853)

4η 0.0000768** (2.044708)

0.181389 (0.94781)

5η 0.0000767** (2.002506)

0.110958 (0.794972)

2R Log likelihood

0.654326 4553.046

1884.558

SIC 18.84171 1844.401 Q(12) 119.19* 14.971 Q2(12) 9.3877 11.229 Skewness 4.0547 -0.11269 Kurtosis 17.695 3.1841 Jarque-Bera 254.250* 1.6979 ARCH(6)-LM 1.37925 1.7506 REST(12,6) 50.2854* 2.3293

Notes:*, **, *** significant at 1%, 5% and 10% respectively.

23

V. A Summary and Conclusions

The recent history of the Hashemite Kingdom of Jordan shows that the country has

passed through many internal and external disturbances (economic and political).

These include the 1967 Arab Israeli War, the 1970 Civil War, the 1973 War, the 1989

devaluation of the Jordanian Dinar and the introduction of the IMF supported

austerity measures, the 1991 Gulf War, the 1994 signing of the peace treaty with

Israel, the 1998 beginning of the late His Majesty King Hussein's (HMKH)

chemotherapy for non-Hodgkin's lymphoma, the 1999 death of HMKH, and more

recently the 2001/2002 Intifada in the Palestinian territories.

It can be argued that the most natural occurrence and peaceful change that Jordan had

to face, the death of King Hussein, was, potentially, the most important as far as its

impact on ASM’s share prices. Indeed, this event in the contemporary history of

Jordan is important for its emotional, political and economic implications. King

Hussein's death was felt far outside his grieving Kingdom.

This paper used the standard GARCH, exponential GARCH (EGARCH) and the GJR

models to examine the conditional volatility associated with two major dates

associated with King Hussein. These are the official denial that Hussein may die

within three months and the date of his death.

The results indicate that King Hussein’s health “rumors” and the date of his death

caused extremely high volatility in the Jordanian capital market. Moreover, the results

suggest that these events are discounted in at least three days following their

respective dates.

24

On January 25, 1999 the late King Hussein of Jordan named his eldest son heir to the

throne. His Majesty King Abdullah attended military school in England and the

United States and served as commander of the Jordanian Army’s special forces. In his

first appearances, he showed much of his father’s charisma. Moreover, it is hoped that

His Majesty King Abdullah (and Jordan) will experience greater political and

economic stability than that faced by the late King Hussein.

25

References Alesina, A. (1987) Macroeconomic policy convergence in a two party system with rational voters, Quarterly Journal of Economics, 102, 651-78. Alesina, A. and Cukierman, A. (1990) The politics ambiguity, Quarterly Journal of Economics, 829-50. Alesina, A. and Perotti, R. (1996) Income distribution, political instability, and investment, European Economic Review, 40, 1203-1228. Alesina, A., Ozler, S., Roubini, N. and Swagel, P. (1996) Political instability and economic growth, Journal of Economic Growth, 1, 189-211. Bekaert G., Erb C. Harvey C. and Viskanta T. (1998) Distributional characteristics of emerging market returns and asset allocation, Journal of Portfolio Management, 24, 102-16. Bekaert, G. and Harvey, C. (1997) Emerging equity market volatility, Journal of Financial Economics, 43, 29-77. Bekaert, G. and Wu, G. (2000) Asymmetric volatility and risk in equity markets, Review of Financial Studies, 13, 1-42. Bittlingmayer, G. (1998) Output, stock volatility, and political uncertainty in a national experiment: Germany, 1880-1940, Journal of finance, 53, 2243-58. Bollersev, T. (1986) Generalized autoregressive conditional heteroscedasticity, Journal of Econometrics, 33, 307-27. Bollerslev, T. and Wooldridge, J. (1992) Quasi-maximum likelihood estimation and inference in dynamic models with time varying covariances, Econometric Reviews, 11, 143-72. Campos, N., Nugent, J. and Robinson, J. (1999) Can political instability be good for growth? The case of the Middle East and North Africa, Working Paper, University of Southern California, mimeo. Caselli, F., Esquivel, G. and Lefort, F. (1996) Reopening the convergence debate: a new look at cross-country empirics, Journal of Economic Growth, 1, 363-90. Cutler, D., Poterba, J. and Summers, L. (1989) What moves stock prices? Journal of Portfolio Management, 15, 4-11. De Long, J., Shleifer, A., Summers, L and Waldmann, R. (1989) The Size and incidence of the losses from noise trading, Journal of Finance, 44, 681-96.

26

De Long, J., Shleifer, A., Summers, and Waldmann, R. (1990) Positive feedback investment strategies and destabilizing rational speculation, Journal of Finance, 45, 379-95. Diba, B. and Grossman, H. (1988) Explosive rational bubbles in stock prices? American Economic Review, 78, 520-30. Diebold, F. (1986) Temporal aggregation of ARCH process and the distribution of assets returns, Special Studies Paper 200, Federal Reserve Board, Division of research and Statistics: Washington DC. Engle, R. (1982) Autoregressive conditional heteroskedasticity with estimates of the variance of UK inflation, Econometrica, 50, 987-1008. Engle, R. and Bollerslev, T. (1986) Modelling the persistence of conditional variance, Econometric Reviews, 5, 1-50. Erb, C., Harvey, C. and Viskanta, T. (1996) Expected returns and volatility in 135 countries, Journal of Portfolio Management, 22, 46-58. Flood, R. and Graber, P. (1984) Collapsing exchange rate regimes: some linear examples, Journal of International Economics, 17, 1-13. Flood, R. and Hodrick, R. (1986) Asset price volatility, bubbles, and process switching, Journal of Finance, 41,831-42. Glen, A., Lee, K. and Singh, A. (2000) Competition, corporate governance and financing of corporate growth in emerging markets, Working Paper, Cambridge University, mimeo. Glosten, L., Jaganathan, R. and Runkle, D. (1993) On the relation between the expected value and volatility of the normal excess return on stocks, Journal of Finance, 48, 1779-1801. Kim, J. and Singal, V. (2000) Stock market of openings: experience of emerging economies, Journal of Business, 73, 25-66. Lamoureux, C. and Lastrapes, W. (1990) Hetrosckedasticity in stock return data: volume versus GARCH effects, Journal of Finance, 45, 221-29. LeRoy, S. and Porter, D. (1981) The present-value relation: tests based on implied variance bounds, Econometrica, 49, 555-74. Levine, R. and Zervos, S. (1998) Stock markets, banks and economic growth, American Economic Review, 88, 539-58. Nelson, D. (1991) Conditional hetroskedasticity in asset returns: a new approach, Econometrica, 59, 347-70.

27

Pagan, A. (1996) The Econometrics of financial markets, Journal of Empirical Finance, 3, 15-102. Rabemanananjara, R. and Zakoian, J. (1993) Threshold ARCH models and asymmetric in volatility, Journal of Applied Econometrics, 8, 31-49. Ramsey, J. (1969) Tests for specification errors in classical least-squares regression analysis, Journal of the Royal Statistical Society, 31, 350-71. Roll, R. (1988) R2, Journal of Finance, 43, 541-66. Schwert, W. (1989) Why does stock market volatility change over time? Journal of Finance, 44, 1115-1153. Shiller, R. (1981) Do stock prices move too much to be justified by subsequent changes in dividends, American Economic Review, 71, 421-36. Singh, A. (1992) The stock market and economic development: should developing countries encourage stock markets? Working Paper No. 49, University of Cambridge. Singh, A. (1997) Financial liberalization, stock markets and economic development, The Economic Journal, 107, 771-82. Singh, A. (1999) Should Africa promotes stock market capitalism? Journal of International Development, 11, 343-65. Singh, A., Singh, A. and Weisse, B. (2000) Information technology, venture capital and the stock market, Working Paper, University of Cambridge, mimeo. Spyrou, S. and Kassimatis, K. (2001) Stock and credit market expansion and economic development in emerging markets: Further evidence utilizing cointegration analysis, Applied Economics, 33, 1057-64. Willard, K, Guinnane, T. and Rosen, H. (1996) Turning points in the civil war: views from the Greenback market, American Economic Review, 86, 1001-18. Zakoian, J. (1994) Threshold hetroscedastic models, Journal of Economic Dynamics and Control, 18, 931-55.